Categorical variable as dep and indep... PreviousNext
Mplus Discussion > Categorical Data Modeling >
Message/Author
 Marie Alice posted on Monday, September 21, 2015 - 1:50 pm
I am having difficulty with a categorical variable that is both a dependent variable and an independent variable.

I am trying to test whether income mediates the relationship between single parent and math achievement, controlling for parent's education on all factors.
More precisely:
Math on singlem singled income educ;
income on singlem singled educ;
singlem on educ;
singled on educ;
(where math=continuous, income=continuous, educ= continuous, singlem, singled=dummy variables, two parent as reference)

But I realize the last two lines would not be accurate, because the reference groups becomes ambiguous.
I get an error message when I make one categorical variable 'single parent'(with three groups) and define it as "nominal=."

Is there any way I could use multi-group analyses in this case, although single parent is not a moderator?
 Bengt O. Muthen posted on Tuesday, September 22, 2015 - 12:41 am
I would delete the lines

singlem on educ;
singled on educ;

because you are not trying to predict who is singlem, singled. This then also avoids the nominal issue.
 Marie Alice posted on Thursday, September 24, 2015 - 12:43 am
Thank you for your quick reply.
However, it is essential that I control for the effect of parent's education on single parent (singlem and singled), as past research has shown that those with less education tend to have higher probability of becoming single parent. My interest is in testing, controlling for that effect (selection bias), to what extent does being a single parent provide disadvantage to children's academic achievement, and to what extent is it mediated by income deprivation?
Is there anyway in modeling the above either via "nominal" or "multi-group analyses"?
 Bengt O. Muthen posted on Thursday, September 24, 2015 - 11:35 pm
It is more complex, but you can use the approach of a nominal mediator explained in the paper on our website under

http://www.statmodel.com/Mediation.shtml

Muthén, B. (2011). Applications of causally defined direct and indirect effects in mediation analysis using SEM in Mplus. Click here to download the paper. Click here to view the Technical appendix that goes with this paper and click here for the Mplus input appendix. Click here to view Mplus inputs, data, and outputs used in this paper.

But note that "controlling for z" often means to just add z to the other covariates so that the effect of a covariate x on y is the effect holding z constant.
Back to top
Add Your Message Here
Post:
Username: Posting Information:
This is a private posting area. Only registered users and moderators may post messages here.
Password:
Options: Enable HTML code in message
Automatically activate URLs in message
Action: